AI Hub
45
Models
23
Providers
21
Vendor
24
Open-source
A reference catalog for orientation. Specifications such as context windows and licenses are indicative and change frequently — always verify with the provider. Click a model for its full profile.
45 models
Kimi K2.7 Code
Moonshot AI (Kimi)
Code model
Agentic coding and software-engineering tasks as an open-weight model from Moonshot AI.
Best for
Llama Nemotron Rerank VL 1B v2
NVIDIA
Reranker
Compact NVIDIA reranking model with vision support for retrieval and RAG pipelines.
Claude Fable 5
Anthropic
Reasoning model
A recent flagship model in the Claude family, positioned as Anthropic's most powerful tier, built for the most demanding reasoning, agentic, and knowledge tasks.
Best for
Nex-N2-Pro
Nex AGI
Language model
Nex AGI's general-purpose large language model.
Riverflow V2.5 Pro
Sourceful
Image generation
Sourceful's image-generation model.
Nemotron 3.5 Content Safety
NVIDIA
Safety model
NVIDIA guard model that classifies inputs and outputs for content-safety moderation.
Nemotron 3 Ultra
NVIDIA
Language model
General-purpose text generation and reasoning as an openly available large model, optimized for NVIDIA GPU inference.
Best for
MAI-Voice-2
Microsoft
Speech model
Microsoft's in-house text-to-speech voice model (MAI line).
MAI-Transcribe 1.5
Microsoft
Speech model
Microsoft's in-house speech-to-text transcription model (MAI line).
MAI-Image-2.5
Microsoft
Image generation
Microsoft's in-house image-generation model (MAI line).
MiniMax M3
MiniMax
Language model
General-purpose tasks with emphasis on very long context understanding as an open large model.
Best for
Step 3.7 Flash
StepFun
Language model
StepFun's fast, cost-efficient general-purpose language model.
Claude Opus 4.8
Anthropic
Reasoning model
Anthropic's flagship model — strong agentic coding, long-horizon reasoning and careful instruction following.
Best for
Parakeet TDT 0.6B v3
NVIDIA
Speech model
NVIDIA's compact, low-latency automatic speech-recognition model.
Qwen3.7 Max
Qwen (Alibaba)
Language model
General-purpose reasoning, coding, and multilingual tasks as Qwen's large proprietary API flagship tier.
Best for
Grok Build 0.1
xAI
Code model
Software building and agentic coding workflows as an xAI Grok variant aimed at developers.
Best for
Gemini Embedding 2
Google DeepMind
Embedding model
Google's text-embedding model for semantic search, retrieval and RAG pipelines.
Gemini 3.5 Flash
Google DeepMind
Multimodal model
Latency- and cost-optimized Gemini variant with long context and multimodal input for high-volume workloads.
Best for
Grok Imagine Video
xAI
Video generation
xAI's text-to-video generation model.
Grok Imagine Image (Quality)
xAI
Image generation
xAI's text-to-image generation model for creating images from prompts.
Voxtral Mini Transcribe
Mistral AI
Speech model
Mistral's open speech-to-text transcription model.
Grok Voice TTS 1.0
xAI
Speech model
xAI's text-to-speech voice model.
Qwen3 ASR Flash
Qwen (Alibaba)
Speech model
Alibaba's fast automatic speech-recognition model in the Qwen family.
Recraft V4.1 Pro
Recraft
Image generation
Recraft's higher-quality professional text-to-image generation model.
Recraft V4.1
Recraft
Image generation
Recraft's text-to-image generation model for design and creative work.
Perceptron Mk1
Perceptron
Multimodal model
Perceptron's multimodal model with vision understanding.
Ring-2.6-1T
inclusionAI
Language model
inclusionAI's large mixture-of-experts language model with long context.
GPT-5
OpenAI
Multimodal model
OpenAI's flagship multimodal model with strong reasoning, tool use and agentic workflows.
Best for
gpt-oss-120b
OpenAI
Language model
OpenAI's open-weight model line for self-hosting under a permissive license.
Best for
Qwen3
Qwen (Alibaba)
Language model
Alibaba's open Qwen family spanning dense and MoE sizes with strong multilingual and coding skills.
Best for
o3
OpenAI
Reasoning model
OpenAI reasoning model that spends inference-time compute on step-by-step problem solving.
Best for
Llama 4
Meta AI
Multimodal model
Meta's open-weight mixture-of-experts family with native multimodality and very long context.
Best for
Gemini 2.5 Pro
Google DeepMind
Multimodal model
Google DeepMind's long-context multimodal flagship handling text, images, audio and video.
Best for
DeepSeek-V3
DeepSeek
Language model
Large MoE model offering frontier-class quality at low inference cost, openly available.
Best for
Gemma 3
Google DeepMind
Multimodal model
Google's lightweight open model derived from Gemini research, runs on a single GPU.
Best for
DeepSeek-R1
DeepSeek
Reasoning model
Open reasoning model trained with large-scale reinforcement learning, with distilled variants.
Best for
Phi-4
Microsoft
Small language model
Microsoft small-language-model line with high reasoning quality per parameter.
Best for
Llama 3.3 70B
Meta AI
Language model
Dense 70B instruction-tuned model; a popular self-hosting baseline.
Best for
Amazon Nova Pro
Amazon
Multimodal model
Amazon's multimodal foundation model on Bedrock, positioned for price-performance.
Best for
Stable Diffusion 3.5
Stability AI
Image generation
Stability AI's open text-to-image family, free for research and below a revenue threshold.
Best for
FLUX.1
Black Forest Labs
Image generation
Black Forest Labs image-generation family; schnell is Apache 2.0, dev is non-commercial.
Best for
Mixtral 8x22B
Mistral AI
Language model
Mistral sparse mixture-of-experts model with strong quality and efficient inference.
Best for
Command R+
Cohere
Language model (RAG)
Cohere's retrieval-augmented model optimized for enterprise RAG and tool use.
Best for
DBRX
Databricks
Language model (MoE)
Databricks' open MoE model aimed at enterprise data and analytics use cases.
Best for
Whisper
OpenAI
Speech model
Widely-used open multilingual speech recognition and translation model.
Best for
